game-theory-lens
Pass
Audited by Gen Agent Trust Hub on May 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions in
SKILL.mdspecify using WebSearch to enrich user-provided information. This processes untrusted external content into the agent's context. - Ingestion points: WebSearch results are ingested during the information gathering phase.
- Boundary markers: Absent. The skill does not provide specific instructions to the model to treat search results as untrusted data or use delimiters.
- Capability inventory: The skill performs local file writes to
{CWD}/game-theory-lens/and uses WebSearch. - Sanitization: Absent.
- [COMMAND_EXECUTION]: The skill is instructed to save analysis output to a specific local directory (
{CWD}/game-theory-lens/). While this is a functional requirement for report generation, it involves interaction with the local file system.
Audit Metadata